When the expansive digital communities made possible by smart contracts and DAOs turn their attentions to actually getting things done, they often run into roadblocks. Executing on their mission — and doing so efficiently — is fraught with difficulty for a DAO which traces its routes back to a distributed community.
Such DAOs generally comprise low influence, high stakes members.
Low influence because there are a lot of them in the community
High stakes because by definition, this is where the problems come in. (Communities in low-stake DAOs are less concerned by inefficient execution, because members have less riding on good execution.)
These are communities where each member has a relatively large chunk of their income, social status, or happiness tied up with the success of the community, yet there are so many of them that their individual influence remains negligible.
Consider Packy McCormick’s overview of Braintrust, a freelance platform owned by the freelancers:
“Let’s say I own $10k worth of Amazon shares and am an Amazon customer, and let’s also say that they put the decision to raise Prime prices $10 to a vote (which they wouldn’t, but play along). That $10 is meaningless to me relative to my shares.
If, however, I own $10k of Braintrust and they propose a 10% fee increase, and I make $72k on Braintrust, then the vote will have a $7,200 impact on me.”
I have successfully used it for hiring writers at Culture3 – shoutout to Steph, I know you’re reading! But the thing for writers like Steph is that whilst you have a big stake in Braintrust as an income source, as one of thousands of members, you have little influence. Service DAOs (similar but different) have similar dynamics.
Braintrust has almost 4,000 freelancers (or addresses), making your vote a drop in the ocean. The smart thing to do here is for groups of freelancers to pool their votes in unison. Over time, individually-uninfluential voting members will coalesce to vote for their interests on their behalf.
This is already happening in the more developed DAOs. The exquisitely-named MonetSupply is a delegate for voters in the Uniswap, Compound, and Aave DAOs. Across these, hundreds have delegated their votes to him, and at Uniswap, where he is the 20th largest holder, he has voted on more proposals than any of the high-influence members above him, indicating the significance of a delegated union. I also checked Compound, and it’s true there too. You can involve people in governance without forcing them to actively get involved.
(As an aside, I think an underrated part of the transparency virtue in DAOs is making it easy to find people with similar interests to you. History is full of influential people masquerading as outsiders, but when their voting stake and history is onchain, it’s trivial to verify who is an outsider and who isn’t.)
Delegation aligns with the natural voting priorities of most low influence-high stake members: vision over detail. Most voters care deeply about the vision, the high-level decisions, the strategy. This is particularly the case when a voter has little influence over anything, because the natural choice is to focus on the big picture things that influence everything else. As outlined above, this lets them influence major decisions affecting users (like fees) and the market (like being on a specific blockchain).
That these members emphasise vision over detail becomes clear when we look at the statistics. Curve DAO (a finance platform) indicates just how little community members care about voting. Just 7% of Curve tokens have been ‘locked’ for voting; the holders of the other 93% have chosen to trade instead and simply benefit from the financial return (a cut of trading fees). Looking more broadly, one analysis looks at almost ten thousand DAOs, who have amassed 5.6 million votes on 63,000 proposals. I believe this data covers the entire history of each DAO, up until last month.
Their findings reveal three main themes:
DAO members value simplicity very highly. 88% of proposals use a single-choice voting system, such as Yes/No. This means that members essentially have a gun to their head on every vote: “approve this specific policy or nothing gets approved.”
There is limited engagement in governance: the average DAO proposal receives 89 votes, when DAO membership tends to be in the low thousands.
Engagement is concentrated in the biggest DAOs. The top 10 DAOs are responsible for 13% of all proposals and 38% of all votes.
In sum: most people hardly engage with their DAOs. This creates governing vacuums when there should be community engagement. The solution is to channel your community towards high leverage areas (like strategy) and the high-visibility areas (like market choices), without burdening it with technical topics. This tends to orientate community around three areas:
Expressing the vision for the organisation
Appealing to customers in the market
Preventing armageddon, like drastic changes to a platform
What does this mean for mass community governance? It means that to be successful, governance needs to be accessible, and that means delegating responsibility, automating accountability, and not being too onerous.
Delegation. Vote delegation is one half of this, and blockchain tools like Tally make it easy to delegate votes on a vote by vote, DAO by DAO basis. Separate systems and rules can be constructed to prevent abuse, including onchain reputation, delegation reviews, and time-limited delegation.
But DAOs differ from traditional organisations not only in having a thousand CEOs rather than one, implying the need for delegated votes. They also differ in having very few employees, whereas a traditional organisation will have thousands.
Managing employees is far too complex for an expansive DAO, because management relies on many inter-connected decisions (or ‘votes’) that a DAO community will not pay enough attention to. Instead, it is better to delegate resources to small groups2 via similar systems and rules through which the community delegates their votes. Nouns DAO, a decentralised brand, is an excellent example. Recent votes have given funds to different groups to execute a video series, a clothing brand, a sponsorship deal, and an advertising campaign. The voters pick the broad vision and find external teams to make it happen. A 271-member DAO, Nouns members each have significant influence; as DAOs grow larger, they will embrace modularity and reserve only more abstract decisions for all members.
Onchain accountability. Andreas Schedler writes that “A is accountable to B when A is obliged to inform B about A’s actions and decisions, to justify them, and to suffer punishment in the case of eventual misconduct” (pg. 18 of pdf download). My addition is that the punishment has to be known to be sufficiently severe by person A before they perform misconduct, otherwise the punishment cannot be a relevant deterrent.
Financial decisions like funding are usually a good fit for this, since money can be deposited in a smart contract as programmable escrow. The catch-22 is that most actions simple enough to be identified as misconduct by code can also be prevented by code without needing to resort to a deterrent. Meanwhile, misconduct that cannot be identified by code may also be too complex to be punished by mass voting (not least when token voting rights can be bought). But I expect there is a middleground: automated accountability can be applied much more than it is today.
You can go too far by decomposing operations into modules and decisions into accessible votes**.** Forming and executing good strategies requires interconnected choices that cannot all be modified away. The risk is that each module and each vote occurs under slightly different interpretations of the DAO’s overarching strategy, that these interpretations become more and more distinct over time, ultimately preventing the DAO as a whole from achieving its goals.
Systems to keep modules and votes aligned are important, but ultimately, these organisations are not going to be as strategic as their more coordinated cousins; and organisations with more complex, strategic goals should not be set up this way. Different DAO structures are best for different things. In an army, you would not decentralise the general, but you might decentralise reconnaissance. Let me explain.
Low influence-high stakes DAOs should all work to mitigate the governance risk of a community executive that pays little attention to details. But at their best, these DAOs embrace the real advantages of their broad communities: finding gems.
The virtue of DAOs like these is not complex strategy. It is that they are accessible to a wide range of people, some of whom will be genuine gems amongst the rough. To bring closure to the army example, Bellingcat is a perfect example. A “distributed, collaborative” investigative group in their own words, Bellingcat conducts investigations using open source information (such as Google Maps) to make valuable intelligence contributions on events like the Skiprals’ poisoning in 2018 and the downing of flight MH17 in Ukraine in 2014.
One reason Bellingcat has been successful is because reconnaissance can be a modular operation in the digital age. Thousands of people with an MI5 ID cannot necessarily do better than an organisation with just 15 full-time staff and 120 contributors. Similarly, Nouns DAO, fundamentally a brand, only needs a handful of people in a studio to make its video series.
Service DAOs are usually a great example of building on this model. The DAO becomes a curated marketplace, bringing individual providers together to do intelligence work or copywriting for clients who have already been signed up ‘centrally’3.
Such organisations already exist outside web3: we call them partnerships, and they’re often lawyers or consultants; i.e. service providers. In both cases, partners or holders have ownership, but in DAOs, contributing is much more accessible. Partnerships are centralised to maintain a high bar for quality, but this means they miss out on some high quality contributors. DAOs take the opposite approach: taking all the high quality contributors, but some low quality ones slip through too.
The challenge for these DAOs is to build institutions that elevate the diamonds in the rough. Or rather, given the mass nature of these communities, create conditions where the diamonds elevate themselves. This is classic signalling theory. What these DAO want is ‘people to do stuff’: to build features, execute ideas, or act as delegates.
The thing is, some are better at this than others, which means these people find it easier to do a good job. What these ‘institutions’ have to achieve is a situation making it easy enough for the good contributors to contribute, but not so easy that everyone else has a go as well. The problem for DAOs is to figure out what ‘good’ means, and how it balances competence (what they do), enthusiasm (how much), and style (how).
To find those who bring competence, giving them opportunity is enough. Cabin DAO operates as a ‘do-ocracy’, in which anyone can be given $200 to ‘do stuff’, if 3 of their 154 members give a literal thumbs up. Watch the do-ers, reward those who do well with social prestige and greater opportunities, introduce them to other do-ers, and watch them take your DAO to new heights. In a world where you can’t trust anyone’s LinkedIn, just give them a chance and see what they can do.
How about enthusiasm and style? This is less cut and dry, but all DAOs look to the same things at the beginning: Discord. Today, you could call a DAO a Discord-arranged organisation. As laughable as that is, Discord is where DAOs are organised. It’s at least a core part, which means that those who are enthusiastic shine through like, well, diamonds in the rough.
Every DAO will have had this experience, and Jenkins the Valet is one of the best examples of it. Jenkins the Valet is a storytelling ecosystem; it’s a place where communities collaborate on telling their own stories. Their Director of Business Operations, known by the pseudonym FilmBook, originally started out as one of a few thousand community members. His engagement in the project was clear to see. It’s a great example of signalling: exhibiting interest is easy to do for someone who is genuinely interested, but tiresome for someone faking it. Style is harder to assess, though it matters just as much. But watch how they do and how they engage, and it will become clear.
In summary, DAOs made up of large numbers of individually-uninfluential people who nonetheless have good reason to care are well-placed to tap the benefits of community-led governance. However, this community breadth is a risk to engagement, even with effective delegation. DAOs that design for this and mitigate the risk of governance vacuums are better-poised to succeed.